Rendering vector data on top of surfaces is a challenge because the surface may not be precisely known to the rendering computing device. For example, if the rendering is provided by a third-party for a particular application. In addition, the surface shape may change depending on the camera view as with level-of-detail terrain or level-of-detail photogrammetry models. The surface shape may also change regardless of level-of-detail such as for time-dynamic surfaces.